feat: added hooks for collection on append

music_kraken/connection/cache.py

@@ -6,6 +6,7 @@ from typing import List, Optional
 from functools import lru_cache
 import logging
 
+from ..utils import output, BColors
 from ..utils.config import main_settings
 from ..utils.string_processing import fit_to_file_system
 
@@ -204,9 +205,12 @@ class Cache:
         for path in self._dir.iterdir():
             if path.is_dir():
                 for file in path.iterdir():
+                    output(f"Deleting file {file}", color=BColors.GREY)
                     file.unlink()
+                output(f"Deleting folder {path}", color=BColors.HEADER)
                 path.rmdir()
             else:
+                output(f"Deleting folder {path}", color=BColors.HEADER)
                 path.unlink()
 
         self.cached_attributes.clear()

music_kraken/objects/artwork.py

@@ -1,18 +1,14 @@
 from __future__ import annotations
 
-from typing import List, Optional, Dict, Tuple, Type, Union, TypedDict
-
-from .collection import Collection
-from .metadata import (
-    Mapping as id3Mapping,
-    ID3Timestamp,
-    Metadata
-)
-from ..utils.string_processing import unify, hash_url
-
-from .parents import OuterProxy as Base
+from typing import Dict, List, Optional, Set, Tuple, Type, TypedDict, Union
 
 from ..utils.config import main_settings
+from ..utils.string_processing import hash_url, unify
+from .collection import Collection
+from .metadata import ID3Timestamp
+from .metadata import Mapping as id3Mapping
+from .metadata import Metadata
+from .parents import OuterProxy as Base
 
 
 class ArtworkVariant(TypedDict):
@@ -23,7 +19,9 @@ class ArtworkVariant(TypedDict):
 
 
 class Artwork:
-    def __init__(self, *variants: List[ArtworkVariant]) -> None:
+    def __init__(self, *variants: List[ArtworkVariant], parent_artworks: Set[Artwork] = None) -> None:
+        self.parent_artworks: Set[Artwork] = parent_artworks or set()
+        
         self._variant_mapping: Dict[str, ArtworkVariant] = {}
 
         for variant in variants:
@@ -44,21 +42,36 @@ class Artwork:
             "deviation": self._calculate_deviation(width, height),
         }
 
+    @property
+    def flat_empty(self) -> bool:
+        return len(self._variant_mapping.keys()) <= 0
+
+    def _get_best_from_list(self, artwork_variants: List[ArtworkVariant]) -> Optional[ArtworkVariant]:
+        return min(artwork_variants, key=lambda x: x["deviation"])
+
     @property
     def best_variant(self) -> ArtworkVariant:
-        if len(self._variant_mapping.keys()) <= 0:
-            return None
-        return min(self._variant_mapping.values(), key=lambda x: x["deviation"])
+        if self.flat_empty:
+            return self._get_best_from_list([parent.best_variant for parent in self.parent_artworks])
+        return self._get_best_from_list(self._variant_mapping.values())
 
     def get_variant_name(self, variant: ArtworkVariant) -> str:
         return f"artwork_{variant['width']}x{variant['height']}_{hash_url(variant['url']).replace('/', '_')}"
 
     def __merge__(self, other: Artwork, **kwargs) -> None:
+        self.parent_artworks.update(other.parent_artworks)
+
         for key, value in other._variant_mapping.items():
             if key not in self._variant_mapping:
                 self._variant_mapping[key] = value
 
+    def __hash__(self) -> int:
+        return id(self)
+
     def __eq__(self, other: Artwork) -> bool:
+        if hash(self) == hash(other):
+            return True
+
         if not isinstance(other, Artwork):
             return False
         return any(a == b for a, b in zip(self._variant_mapping.keys(), other._variant_mapping.keys()))

music_kraken/objects/collection.py

@@ -1,16 +1,44 @@
 from __future__ import annotations
 
-from collections import defaultdict
-from typing import TypeVar, Generic, Dict, Optional, Iterable, List, Iterator, Tuple, Generator, Union, Any, Set
 import copy
+from collections import defaultdict
+from dataclasses import dataclass
+from typing import (Any, Callable, Dict, Generator, Generic, Iterable,
+                    Iterator, List, Optional, Set, Tuple, TypeVar, Union)
 
-from .parents import OuterProxy
-from ..utils import object_trace
-from ..utils import output, BColors
+from ..utils import BColors, object_trace, output
+from .parents import InnerData, OuterProxy
 
 T = TypeVar('T', bound=OuterProxy)
 
 
+@dataclass
+class AppendHookArguments:
+    """
+    This class is used to store the arguments for the append hook.
+    The best explanation is with an examples:
+    
+    ```
+    # this is the action that triggers the append hook
+    album = Album()
+    song = Song()
+    album.song_collection.append(song)
+    ```
+
+    In this case, the append hook is triggered with the following arguments:
+    ```
+    AppendHookArguments(
+        collection=album.song_collection,
+        new_object=song,
+        collection_root_objects=[album]
+    )
+    ```
+    """
+
+    collection: Collection
+    new_object: T
+    collection_root_objects: Set[InnerData]
+
 class Collection(Generic[T]):
     __is_collection__ = True
 
@@ -27,6 +55,7 @@ class Collection(Generic[T]):
             sync_on_append: Dict[str, Collection] = None,
             append_object_to_attribute: Dict[str, T] = None,
             extend_object_to_attribute: Dict[str, Collection] = None,
+            append_callbacks: List[Callable[[AppendHookArguments], None]] = None,
     ) -> None:
         self._collection_for: dict = dict()
 
@@ -41,6 +70,7 @@ class Collection(Generic[T]):
         self.sync_on_append: Dict[str, Collection] = sync_on_append or {}
         self.pull_from: List[Collection] = []
         self.push_to: List[Collection] = []
+        self.append_callbacks: List[Callable[[AppendHookArguments], None]] = append_callbacks or []
 
         # This is to cleanly unmap previously mapped items by their id
         self._indexed_from_id: Dict[int, Dict[str, Any]] = defaultdict(dict)
@@ -141,6 +171,14 @@ class Collection(Generic[T]):
         for attribute, new_object in self.append_object_to_attribute.items():
             other.__getattribute__(attribute).append(new_object, **kwargs)
 
+        append_hook_args = AppendHookArguments(
+            collection=self,
+            new_object=other,
+            collection_root_objects=self._collection_for.keys(),
+        )
+        for callback in self.append_callbacks:
+            callback(append_hook_args)
+
     def append(self, other: Optional[T], **kwargs):
         """
         If an object, that represents the same entity exists in a relevant collection,

music_kraken/objects/song.py

@@ -477,6 +477,8 @@ class Artist(Base):
     general_genre: str
     unformatted_location: str
 
+    artwork: List[Artwork]
+
     source_collection: SourceCollection
     contact_collection: Collection[Contact]
 
@@ -493,6 +495,8 @@ class Artist(Base):
         "lyrical_themes": list,
         "general_genre": lambda: "",
 
+        "artwork": list,
+
         "source_collection": SourceCollection,
         "album_collection": Collection,
         "contact_collection": Collection,
@@ -511,6 +515,7 @@ class Artist(Base):
         notes: FormattedText = None, 
         lyrical_themes: List[str] = None,
         general_genre: str = None, 
+        artwork: List[Artwork] = None,
         unformatted_location: str = None, 
         source_list: List[Source] = None,
         contact_list: List[Contact] = None, 

music_kraken/pages/genius.py

@@ -134,7 +134,7 @@ class Genius(Page):
                 source_list=[source],
                 artist_list=[self.parse_api_object(data.get("artist"))],
                 artwork=artwork,
-                date=ID3Timestamp(**data.get("release_date_components", {})),
+                date=ID3Timestamp(**(data.get("release_date_components") or {})),
             )
 
         if object_type == "song":

music_kraken/pages/musify.py

@@ -457,17 +457,17 @@ class Musify(Page):
         for album_info in soup.find_all("ul", {"class": "album-info"}):
             list_element: BeautifulSoup = album_info.find("li")
             
-            if list_element is not None:
-                artist_soup: BeautifulSoup
-                for artist_soup in list_element.find_all("a"):
-                    artist_source_list = []
-                    href = artist_soup["href"]
-                    if href is not None:
-                        artist_source_list = [Source(self.SOURCE_TYPE, self.HOST + href)]
-                    artist_list.append(Artist(
-                        name=artist_soup.text.strip(),
-                        source_list=artist_source_list
-                    ))
+        if list_element is not None:
+            artist_soup: BeautifulSoup
+            for artist_soup in list_element.find_all("a"):
+                artist_source_list = []
+                href = artist_soup["href"]
+                if href is not None:
+                    artist_source_list = [Source(self.SOURCE_TYPE, self.HOST + href)]
+                artist_list.append(Artist(
+                    name=artist_soup.text.strip(),
+                    source_list=artist_source_list
+                ))
     
         # breadcrums
         breadcrumb_list_element_list: List[BeautifulSoup] = soup.find_all("ol", {"class": "breadcrumb"})
@@ -485,7 +485,7 @@ class Musify(Page):
             
             track_name = list_points[4].text.strip()
 
-        # artwork
+        # album artwork
         artwork: Artwork = Artwork()
         album_image_element_list: List[BeautifulSoup] = soup.find_all("img", {"class": "album-img"})
         for album_image_element in album_image_element_list:
@@ -918,7 +918,8 @@ class Musify(Page):
             name=name,
             country=country,
             source_list=source_list,
-            notes=notes
+            notes=notes,
+            artwork=self._fetch_artist_artwork(soup, **kwargs)
         )
 
     def _parse_album_card(self, album_card: BeautifulSoup, artist_name: str = None, **kwargs) -> Album:
@@ -1056,6 +1057,20 @@ class Musify(Page):
 
             artist.album_collection.append(album)
 
+    def _fetch_artist_artwork(self, soup: BeautifulSoup, **kwargs):
+        # artist artwork
+        artist_artwork: List[Artwork] = Artwork()
+        artist_a_element_list: List[BeautifulSoup] = soup.find_all("a")
+        for artist_a_element in artist_a_element_list:
+            if artist_a_element.find_all("img", {"class": "artist-img"}).count() > 0:
+                artwork_gallery = self.connection.get(artist_a_element("data-src", artist_a_element.get("href")))
+                if artwork_gallery is not None:
+                    gallery_image_element_list: List[BeautifulSoup] = artwork_gallery.find_all("img", {"class": "artist-img"})
+                    for gallery_image_element in gallery_image_element_list:
+                        artist_artwork.push(Artwork(url=gallery_image_element.get("data-src", gallery_image_element.get("src"))))
+
+        return artist_artwork
+
     def fetch_artist(self, source: Source, **kwargs) -> Artist:
         """
         TODO
@@ -1068,7 +1083,7 @@ class Musify(Page):
 
         artist = self._fetch_initial_artist(url, source=source, **kwargs)
         self._fetch_artist_discography(artist, url, artist.name, **kwargs)
-        
+        self._fetch_artist_artwork(artist, **kwargs)
         return artist
 
     def fetch_label(self, source: Source, stop_at_level: int = 1) -> Label:

music_kraken/pages/youtube_music/youtube_music.py

@@ -549,6 +549,11 @@ class YoutubeMusic(SuperYouTube):
         return album
 
     def fetch_lyrics(self, video_id: str, playlist_id: str = None) -> str:
+        """
+        1. fetches the tabs of a song, to get the browse id
+        2. finds the browse id of the lyrics
+        3. fetches the lyrics with the browse id
+        """
         request_data = {
             "context": {**self.credentials.context, "adSignalsInfo": {"params": []}},
             "videoId": video_id,
@@ -575,7 +580,8 @@ class YoutubeMusic(SuperYouTube):
             pageType = traverse_json_path(tab, "tabRenderer.endpoint.browseEndpoint.browseEndpointContextSupportedConfigs.browseEndpointContextMusicConfig.pageType", default="")
             if pageType in ("MUSIC_TAB_TYPE_LYRICS", "MUSIC_PAGE_TYPE_TRACK_LYRICS") or "lyrics" in pageType.lower():
                 browse_id = traverse_json_path(tab, "tabRenderer.endpoint.browseEndpoint.browseId", default=None)
-                break
+                if browse_id is not None:
+                    break
 
         if browse_id is None:
             return None
@@ -721,7 +727,6 @@ class YoutubeMusic(SuperYouTube):
 
         self.download_values_by_url[source.url] = {
             "url": _best_format.get("url"),
-            "chunk_size": _best_format.get("downloader_options", {}).get("http_chunk_size", main_settings["chunk_size"]),
             "headers": _best_format.get("http_headers", {}),
         }